Automation is often seen as destroying jobs, but new Wharton research shows it also can quietly block workers from moving into better-paid roles.
Industrial robots do not just shape pay today. Research from Pinar Yildirim, a Wharton professor of economics and marketing, shows they also make workers less likely to move into higher-paying occupations, cutting expected lifetime earnings.
“Workers aren’t necessarily losing their jobs, and it doesn’t feel like a sudden shock,” Yildirim said. “It’s more that the upward moves become a bit less likely, and over time this aggregates into careers that are more stagnant and less likely to end up in well-paying senior positions.”
According to the working paper, and an article published by the Brookings Institution based on the study, one additional robot per 1,000 workers reduces expected lifetime earnings by about 1.5% (roughly $3,360 in today’s money). About a third of the decline comes not from lower pay within jobs, but from workers becoming less likely to move into better-paid ones in future.
Automation, in other words, is upending careers as much as roles. How career mobility is quietly deteriorating
The pattern has been building for years; the paper draws on earlier work showing a decline in occupational mobility across the U.S., with workers becoming less likely to move into higher-earning jobs between 2000 and 2017.
That shift shows up in how people behave, according to the more recent paper. Areas where workers had stronger long-term earning prospects saw higher levels of college education enrollment and homebuilding, suggesting that when opportunities to improve earnings weaken, people may pull back from investing in their futures.
Better opportunities to move up at work were linked to roughly 23% more homebuilding and a 1.1 percentage point increase in college education.
More broadly, the labour market can look stronger than it really is, with the paper finding that career mobility has deteriorated even as real wages were still rising. From 2000 to 2016, workers were still seeing gains from higher wages - but were becoming less likely to move into higher-paying jobs.
Rising wages added about $16,100 to expected lifetime earnings over that period, while weaker career progression wiped out almost $12,500 of those gains.
That helps explain why headline economic indicators such as unemployment can be misleading, the paper suggests.
The effects are not confined to one group. A university degree did not shield people from industrial automation: Areas with greater exposure to robots saw similar declines in career prospects across education levels. The impact was strongest in parts of the U.S. with large manufacturing bases.
Mid-career workers are particularly exposed. Those with between six and 20 years of experience saw the sharpest declines in career prospects, suggesting that people who have spent years building specialised skills can find that those paths start to narrow.
“In practice, declines in career progression often look like declines in mobility to a missing middle rung of the career ladder,” Yildirim said, pointing to fewer moves from junior roles into supervisory and management positions.
Why U.S. labour policy needs to shift
The study itself focuses on industrial robots, examining a period largely before generative AI tools such as ChatGPT came out. It draws on more than 18 million resumes between 2000 and 2017, alongside wage data and measures of where robots were being adopted.
Even so, the logic may extend further. If AI follows a similar path, the effects could spread well beyond the jobs directly exposed, as workers pushed out of some roles compete for opportunities elsewhere in the economy. That matters as companies from Tesla to Amazon deploy robots across factories and warehouses to cut costs and boost productivity.
The consequences are not only economic. Where long-term earning prospects deteriorated, support for Donald Trump rose, with a one standard deviation decline linked to a 0.67 percentage point increase in his vote share. The researchers say the results fit a broader pattern in which economic insecurity and fading expectations about the future boost support for populist politicians.
Taken together, the research points to a labor market where the problem is not just job loss, but the quiet disappearance of pathways to better work.
For policymakers, that distinction matters. U.S. labour policy still focuses heavily on unemployment, overlooking workers whose prospects are quietly eroding even as they remain employed. “Policies targeting only displaced manufacturing workers will miss the majority of affected workers,” the authors write.
“Many existing policy programmes trigger only once someone has actually lost their job. However, if the impact comes in the form of career stagnation, these programmes are not helpful,” Yildirim said.
Instead, the researchers say policy focus needs to shift. Preserving jobs at all costs can trap workers in declining roles rather than helping them move into better long-term opportunities. So, more emphasis is needed on retraining and career mobility.